"This is, as far as I know, the
largest CTDB conversion effort ever attempted with both
terrain and culture. I think there have been other tries,
but none successful."
In today's military environment, utilizing real equipment
to train for war is prohibitively expensive and
environmentally controversial: limitations that translate
into reduced combat readiness. Modern trends towards team
training cannot be met exclusively through using simulators,
which are too costly to be used in any great quantity. A
valuable training scenario needs to realistically recreate
the number, as well as the nature, of combatants. So,
Distributed Interactive Simulation usually requires realtime
war-gaming systems that provide both enemy and friendly
forces. The interoperability of virtual, semi-automated
forces and simulated forces requires that the war-gaming
program and the "man-in-the-loop" simulators operate in the
same environment.
However, simulator systems with visual simulation and
war-gaming components require data in different database
formats that exactly correlate with each other. MultiGen
Inc., the premier supplier of realtime 3D tools for visual
simulation, found its customers asking for commercially
available tools that would enable them to create correlated
databases. MultiGen set out to develop database conversion
tools that would meet those customer needs.
"Knowing what our customers wanted, we took special care
over the design of the architecture of the software," said
Carl Suttle, MultiGen Product Manager. "The result was that
even though this was the first conversion carried out by a
customer, and was the biggest ever attempted, it was
successful."
MODULAR SEMI-AUTOMATED
FORCES (ModSAF)
Loral Advanced Distributed Simulation developed the
ModSAF system for the U.S. Army Simulation, Training, and
Instrumentation Command (STRICOM) and the Advanced Research
Projects Agency - Advanced Systems Technology Office
(ARPA-ASTO). ModSAF is a software architecture that enables
users to create warfighting exercises on either a single
computer or a network of computers. Through its Graphical
User Interface (GUI), ModSAF makes it possible to create
virtual vehicles û or entities û and simulate their
behaviors, change their reactions, monitor status, and
simulate complex battlefield environments within Distributed
Interactive Simulation (DIS) live and virtual simulation
exercises.
ModSAF entities include fixed and rotary wing aircraft,
ground vehicles, and dismounted infantry. Entities behave
with "intelligence": that is, they independently move, fire,
sense, communicate, and react, without operator
intervention.
|
|
|
ModSAF display with GUI, showing
a rural section of MultiGen's Cody, Wyoming,
database.
|
At the vehicle and weapons system level, entities perform
a realistic range of basic actions in keeping with their
type. Simulated tanks drive along roads, scan their turrets,
and turn in place. In addition, entities are able to detect
environmental conditions. For example, a simulated tank
under ModSAF's control will "recognize" a body of water as
impassable terrain and find a route around it, rather than
drive through it. Simulated aircraft ascend, orbit, and land
appropriately, and weapon systems demonstrate correct rates
of fire, and realistic trajectories.
ModSAF simulated entities also exhibit mobility,
firepower, and catastrophic combat damage when hit by enemy
fire. Both fuel and ammunition resources are tracked and
deplete accurately as they use them. Other simulated
capabilities include inter-visibility û the entities "see"
and react to each other û target detection, identification,
and selection; fire planning, and collision detection. These
capabilities are based upon appropriated factors such as
range, motion, activity, visibility, direction, orders, and
threat evaluation.
ModSAF also enables Command and Control behaviors. When a
unit is simulated, ModSAF creates the entities (such as
simulated aircraft or tanks) in the unit, and builds a
structure corresponding to the unit hierarchy. Commands can
then be issued to either the top-level units, or to their
subordinate units or vehicles. ModSAF interprets the
commands and generates unit and vehicle behavior and tactics
accordingly.
ModSAF is a data-driven application, and relies on a
specific database of information to function as described: a
compact terrain database (CTDB). A CTDB database fulfills
ModSAF's data needs. The CTDB database enables terrain
reasoning; so when ModSAF commands a tank to climb a slope
and executes the maneuver, its counterpart in the correlated
visual database also climbs the slope.
ModSAF and CTDB
Overview
CTDB databases and visual databases are usually created
using the same geographical source data, derived from
standard NIMA products or commercial products like Arc Info.
However, visual and CTDB databases differ according to their
intended use.
A visual database contains the geometry that represents
terrain and cultural features, along with other data such as
color and texture for use in the visual rendering. Typically
ModSAF or other Semi-Automated Forces (SAF) systems need the
same terrain and cultural feature geometry, but also need to
be told directly the type of object represented by the
geometry û for example, whether a terrain polygon is made
out of packed earth or water. And because SAFs also need to
reason tactically, and decide how to move entities from one
location to another, they also need a transportation system
network to be provided. This comes from linear data like
roads, paths, and railroads.
Whatever the variety of ModSAF û marine, navy, air
force, or army û it needs a CTDB database, fully correlated
with the visual database, to perform efficient terrain
reasoning and coherent visual simulation. Traditionally,
converting a realtime 3D visual database to a terrain
reasoning CTDB database involved the use of non-commercially
available S1000 modeling tools.
In response to customer requests for COTS tools that
would enable customers to carry out their own CTDB
conversions, MultiGen has developed a converter that
produces correlated CTDB databases directly from OpenFlight
visual databases. It combines terrain, buildings, trees,
digital elevation data, soil type data, and roads, rivers,
railways and other GIS data into a CTDB Version 7 format
file for immediate use in ModSAF.
Cody - Proof Of Concept
In preparation for I/ITSEC in December of 1996, MultiGen
and Texas Instruments carried out a joint project to
demonstrate that correlated CTDB databases could be produced
from OpenFlight visual databases. The demonstration involved
the production of a database of the Cody, Wyoming area, for
a ground warfare
|
|
|
Part of the Cody database,
showing correlation between OpenFlight visual
database and ModSAF.
|
simulation application. It featured correlated terrain
skin, line of sight correlation, and collision detection.
The entities, controlled by ModSAF V 2.1, recognized and
understood obstacles to movement and vision û for example,
simulated tanks "understood" they could drive through
buildings or tree canopies, or see through them. Adversaries
did not open fire unless they could detect an enemy
presence, in this case visually. Ground vehicles followed
the terrain surface without selecting ground clamping.
Having proved the concept, MultiGen Inc. then moved on to
look for a partner to fully test and develop its fledgling
converter.
ADVANCED DISTRIBUTED
SIMULATION TECHNOLOGY (ADST II)
ADST II is a major U.S. Army Simulation, Training and
Instrumentation Command (STRICOM) contract. It requires a
wide range of professional and technical disciplines
supporting the advancement of DIS technologies. The primary
contractor is Lockheed Martin Information Systems Company
(LMIS) in Orlando, Florida. They team with a number of
companies and organizations to provide support in areas such
as engineering analysis, design and integration; logistics;
program management; and operations and maintenance services.
The Lockheed Martin team provides support for the ADST II
Orlando Base Facility and three Core DIS Facilities (CDFs).
The CDFs primarily support experimentation efforts,
including a variety of DIS applications in Combat, Training,
and Material Development; Test and Evaluation, and Dual Use
technologies. Technologies involved include Local and Wide
Area Networks; DIS protocols; virtual, constructive, and
live simulations; virtual prototyping; and computer
generated forces.
In the fall of 1996, ADST II was contracted to support an
USSOCOM project for the Air Force that integrated a
classified large area terrain database with an AC-130U
Testbed for navigator and fire control officer training. An
essential part of the exercise was converting the visual
simulation database into a correlated CTDB database, a
larger conversion than any previously attempted.
The AC-130U is the latest in and most advanced of a long
line of gunships, adopting the nickname "Spooky" from the
original gunship, the AC-47. The U-model all-weather gunship
is the most complex aircraft weapon system in the world,
with more than 609,000 lines of software built into its
mission computers and avionics systems. Its technology
payload also includes an entirely new fire control system
and night and adverse weather capability via all light-level
television, infrared sensors, and the Hughes APQ-180 radar.
This latest sensor technology fully updates the combat
effectiveness of the AC-130U, despite its airframe being the
original Lockheed C-130 design. Electronic counter-measures
have been installed as part of its modern defense system.
The AC-130U is armed with a 25mm Gatling-gun cannon, a
single-barrel, rapid-fire 40mm cannon, and a 105mm Howitzer.
The 4th Special Operations Squadron is one of eight flying
squadrons within the 16th Special Operations Wing at
Hurlburt Field, Florida. The squadron currently operates
thirteen AC-130Us in support of conventional and joint
special forces missions.
MULTIGEN'S ROLE
In June 1997, MultiGen began working with ADST II.
MultiGen had already proved the viability of producing a
correlated CTDB database from an OpenFlight visual database
with the Cody, Wyoming demonstration. Following the success
of the Cody database conversion, ADST II began attempting
their own CTDB conversion, using MultiGen's CTDB converter.
Because it was still under development at MultiGen, the
Solution Center stepped in, both as guide, and to develop
tools and processes to solve problems as they cropped up.
Brian Mannlein, Manager, Technical Support, and Jim
Stenstrom, of MultiGen's technical staff, worked closely
with ADST II throughout the project.
The specific project involved creating a CTDB database,
correlated with geo-specific, large area terrain visual
databases built by Lockheed Martin in Compu-Scene®
TARGET format. Correlation needed to be exact across the
board; for example, every tree had to be in the right place.
Cultural features and models had to stay on the terrain, and
entities needed to "understand" the environment, i.e.
recognizing whether the terrain was mud, water, or a road,
and behave appropriately.
MultiGen's mission was to develop commercially-available
tools that would enable customers to convert visual
simulation databases to correlated CTDB databases without
outside involvement. This is especially important for
military purposes, where much of the data may be classified.
While working on converting the four-cell, South West USA
visual database, Stenstrom developed tools to resolve
problems as engineers encountered them. After creating a
script to help automate the conversion, Mannlein worked with
ADST II and STRICOM engineers by making the script more
user-friendly and training them on the conversion process.
This meant they were able to carry out the next project --
converting a classified six-cell visual database to a
correlated CTDB database -- with minimal hands-on
involvement from MultiGen technical support.
THE DEVELOPMENT
PROCESS
"At the beginning of this project, MultiGen had been
successful at converting small databases to CTDB," Mannlein
said. "The Cody database was 5min X 5mins. We were now
beginning work on a database 48 times that size, or 2
degrees by 2 degrees. We were going into the unknown: no one
had ever created a CTDB database of this magnitude. The
database was comprised of four geo cells that each contained
225 terrain files, as well as countless references to tree
and building files. The major obstacle was the amount of
data that had to be converted. OpenFlight data contains much
more information than is needed in ModSAF and therefore had
to be reduced."
The substantial amount of cultural as well as terrain
data presented another challenge that had to be resolved
early. Each geo cell contained 900,000 trees, all
geo-specific, along with features including roads,
railroads, sandy desert, lakes, and rivers.
|
|
|
Cody, Wyoming, visual and ModSAF
databases, with buildings, tree canopies, water and
airfield.
|
"When we started, we were faced with a very large amount
of terrain data, and had to figure out how to organize it."
said Stenstrom. "Initially we created one big Flight file,
but found that running the file required more than the
available memory. We tried to manage the cultural data by
creating huge temporary files, but the compiler couldn't
really handle those. This became an issue when the tree file
grew to 220 MB. We couldn't even open it, so we needed to
find an alternative way to organize the data."
The solution was twofold -- removing unnecessary
attributes, and tiling. Because ModSAF is a paging system,
and tiles are processed individually, this method of
managing the data proved viable. Stenstrom wrote a program
that tiles all data into manageable sizes. During the
process tree canopies also had to be tiled to accommodate
ModSAF limitations that restricted the maximum number of
vertices per canopy to 5,000.
Stenstrom and Mannlein hit further problems caused by the
complexity of the data and ModSAF's limitations on handling
it. Some areas of the data were extremely complex; in
particular the airfield, with detail down to individual
runway markings that in some cases were less than six inches
long. An image generator usually draws layered polygonal
data, only needing to know that the runway markings are
attributed in the hierarchy as children of the single
polygon runway. However, ModSAF needs to know the
transportability factor of every polygon, and can't
understand the parent/child layering.
Because some IGs cannot draw layered databases
efficiently, MultiGen II Pro already contained the facility
to convert layered databases into a single layer û
MultiGen's unique cut-out sub-faces tool. However, although
this functions by enabling the IG to limit rendering to the
only visible layer, it does so by increasing the number of
polygons. This increased polygonal density of the database
proved to be beyond ModSAF's handling capacity. Mannlein and
Stenstrom solved this problem by "stripping" the data.
"All that ModSAF needed to know for the airfield was what
kind of soil type it was," said Mannlein. "Therefore we
removed the unnecessary polygons from the database before
cutting out sub-faces, which reduced the unused polygonal
densities that may cause compiler failure."
During the Cody project, the database was converted using
a step by step process, utilizing many command line
executables one at a time. Applying this manual process to a
much larger database conversion became tedious and unwieldy.
During the work on the four cell proof of concept database,
Mannlein created a script that combined all the individual
executables into one continuous operation. While working on
site with ADST II in Orlando, Mannlein and Kirk Thomas of
CDI Managed Information Services streamlined the process
further by creating a control file that contains all the
variables needed to convert any database. Once the user has
defined those variables, the control file is "fed" to the
script, and the conversion process is completely automated.
Following the joint project work, the four-cell converted
CTDB database was delivered to the AC-130U program in
October, correlated and running in ModSAF. The project team
succeeded in preparing the ground for ADST II to embark on
converting a classified six cell visual database, with
significantly more culture, as part of the AC-130U Nav/FCO
Testbed project.
CORRELATED CTDB DATABASE
FROM CLASSIFIED VISUAL DATA
In October 1997, ADST II moved directly from the
successful conversion of the four cell visual database to
converting a six cell visual database to CTDB, using the
beta release of MultiGen's CTDB converter. Whereas MultiGen
engineers had been able to work directly with ADST II on the
proof of concept database, the ADST II team was now on its
own. MultiGen could not get involved hands-on because of the
classified data. The terrain covers an area of Florida,
Alabama, and Louisiana, and simulates the training area for
the crews of the AC-130U. Not only was it large area, but
the visual database had much greater cultural density than
the South West USA database. Each cell had 225 OpenFlight
files of multiple LOD visual database. ADST II created the
correlated CTDB database from the highest LOD.
"This is, as far as I know, the largest conversion effort
ever attempted with both terrain and culture," said AC-130U
project manager John Little, of Lockheed Martin. "I think
there have been other tries, but none successful. We found
some problems with ModSAF due to the size of the database,
none of which have required more than a different conversion
approach or more memory in the system that is executing.
CONCLUSION
MultiGen's commercially available CTDB conversions tools
enable the customer to go straight from the visual format,
and the actual polygonal terrain skin, to the terrain
reasoning format. Correlation is exact between the visual
database and the CTDB database. MultiGen II Pro can
regenerate a database many times if required, using
different parameters to meet the needs of different image
generators. MultiGen's CTDB converter follows the guiding
principle of all MultiGen's tools and solutions: ease of
use, accessibility, and ever-improving productivity for the
customer.
Because of the large areas of interest of today's
simulators, the amount of data used by the visual and ModSAF
systems is vast. Creating them requires intelligent and
automatic database generation systems. MultiGen is capable
of creating dense visual large area terrain and CTDB
databases for ModSAF that correlate exactly. And ModSAF is
just one Semi-Automated Forces system. MultiGen believes it
has the tool base and the technical knowledge to support any
customer requirement.
ADDITIONAL INFORMATION
Advanced Training at MultiGen's Solution Center
The Solution Center offers an array of advanced training
programs, and is currently preparing sessions designed to
educate customers in using MultiGen's CTDB conversion tools.
The course will cover the whole process of conversion from a
visual OpenFlight database to CTDB, and incorporate a
multitude of trouble-shooting techniques for handing
potential problems. The sessions will benefit customers who
already have substantial visual simulation experience and
familiarity with MultiGen tools.
üüüüüüü
MultiGen Inc. thanks the individuals and organizations
that contributed to this paper.
MultiGen and OpenFlight are registered trademarks are
registered trademarks of MultiGen Inc. All other trademarks
are the property of their respective owners.
© 1997 MultiGen Inc.
